topk_agg

Description

Returns an approximation of the top K most frequent values in the input, along with their approximate frequencies.

The output is an array of arrays. In the inner arrays, the first entry is the value in the input, while the second entry is its frequency. The outer array contains k elements, sorted by their frequencies in descending order.

Return type

array

Domain

This is an aggregate function (aggregates rows over a group in aggregate verbs.)

This is a window function (calculates over a group of multiple input rows using windowing.)

Categories

• Aggregate

• Semistructured

• Window

Usage

topk_agg(expr, k)

Argument	Type	Optional	Repeatable	Restrictions
expr	variant	no	no	none
k	int64	no	no	constant

Examples

statsby top_names:topk_agg(name, 2), group_by(class)

Given the following input:

name	class
Jack	A
Joe	A
Alice	A
Alice	A
Tom	B
Joe	B
Kathy	B
Mike	A
Tom	B

It shall return the following output:

class	top_names
A	[[“Alice”, 2], [“Jack”, 1]]
B	[[“Tom”, 2], [“Kathy”, 1]]

Note that if there is a tie in the last position the result can be non-deterministic. Any of the values with the same frequency may be included in the last position.